PROJECT SUMMARY Ischemic stroke is the most common cause of disability in US, with motor impairment being the most common form of disability. Rehabilitation therapies remain a mainstay treatment during recovery, but are not very effective for those with moderate to severe upper extremity weakness. Physical and pharmacologic interventions could become more effective if recovery mechanisms were better understood. The studies proposed here will use cutting-edge techniques to dissect the mechanisms of post-stroke recovery at the level of cortical neuronal networks. The principal investigator (PI) is a board-certified neurologist, and has extensive training in neuroscience, stroke neurobiology, and clinical management of patients with neurological diseases. The long- term goal is to understand relationships between the cortical neural network and the motor behaviors with a focus on motor recovery. The proposed research and career development plans will provide the necessary training, mentorship and experience to launch the PI?s career as a successful, R01-funded clinician-scientist. The PI has recently developed deep learning (a branch of machine learning) tools for automated and unbiased analysis of animal behavioral imaging data. Preliminary data demonstrate that, by using these automated tools, 3D kinematics of paw movements can be obtained during a food pellet reaching task in a head-fixed mouse, distinct neural activity patterns emerge in the primary motor cortex (M1) during motor learning of the skilled reaching, and stroke induces changes in the kinematics of the reaching behavior. In the proposed studies, the following specific hypotheses will be tested: stable neuronal network activity patterns emerge in M1 during motor learning (Aim-1); these M1 activity patterns destabilize after a stroke and sensorimotor disconnection (Aim-2); and new M1 activity patterns emerge after post-stroke task-specific rehabilitation (Aim-3). The PI will use, and be further trained on state-of-the-art techniques to record neural population activity (two-photon calcium imaging), to label and silence projection neurons (retro-AAVs and designer-receptors exclusively activated by designer drugs), and to analyze the kinematics of the behavior (deep learning tools). For this purpose, he has assembled a strong mentorship team with experts in their respective fields: Primary mentor Dr. Peyman Golshani (two-photon calcium imaging and analysis in behaving animals), as well as co-mentors, Dr. S. Thomas Carmichael (stroke recovery research), Dr. Jonathan Kao (neural dynamics analysis in motor cortex), and clinically, Dr. Bruce Dobkin (translational neurorehabilitation). The mentored research will be complemented with course work on state-space methods, biostatistics, bioethics, deep learning, responsible conduct of research, and grant writing; as well as improvement of professional skills by publishing manuscripts, presenting at meetings, and participating in workshops on leadership and management. By the end of K08 award, it is expected that the PI will obtain R01 funding as a clinician-scientist and independent investigator, focused on understanding mechanisms of the neural networks that can be modulated to improve motor behavior after stroke.